Multimedia-on-demand (MOD) has grown dramatically in popularity, especially in the domains of education, business, and entertainment. Therefore, the investigation of various alternatives to improve the performance of MOD servers has become a major research focus. The performance of these servers can be enhanced significantly by servicing multiple requests from a common set of resources. The exploited degrees of resource sharing depend greatly on how servers schedule the waiting requests. By scheduling the requests intelligently, a server can support more concurrent customers and can reduce their waiting times for service. In this paper we provide a detailed analysis of existing scheduling policies and propose two new policies, called quantized first-come-first-served (QFCFS) and enhanced minimum idling maximum loss (IML+). We demonstrate the effectiveness of these policies through simulation and show that they suit different patterns of customer waiting tolerance.